Is gemma-4-12B-it-qat-q4_0-unquantized or dolphin-2.9.1-yi-1.5-34b smaller?

Both publish quantizations; compare the table for the exact sizes.

Which uses less memory at long context?

gemma-4-12B-it-qat-q4_0-unquantized — its KV cache at 32K context is 3.0x smaller, because of how its attention layers cache.

Model comparison

gemma-4-12B-it-qat-q4_0-unquantized vs dolphin-2.9.1-yi-1.5-34b

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: gemma-4-12B-it-qat-q4_0-unquantized's KV cache at 32K is 3.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	gemma-4-12B-it-qat-q4_0-unquantized	dolphin-2.9.1-yi-1.5-34b
Parameters	12.0B	34.4B
Architecture	gemma4	llama
Layers	48	60
Native context	262,144	8,192
Mixture of experts	no	no
Quantizations published	2	35
Smallest quantization	6.50 GiB	6.98 GiB
Q4_K_M	—	19.24 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	gemma-4-12B-it-qat-q4_0-unquantized	dolphin-2.9.1-yi-1.5-34b	Ratio
4,096	0.72 GiB	0.94 GiB	1.30×
8,192	0.97 GiB	1.88 GiB	1.94×
16,384	1.47 GiB	3.75 GiB	2.55×
32,768	2.47 GiB	7.50 GiB	3.04×
65,536	4.47 GiB	15.00 GiB	3.36×
131,072	8.47 GiB	30.00 GiB	3.54×